Picture decoding method and apparatus

Abstract

An MPEG downdecoder is to be provided which eliminates dephasing of pixels of moving picture data without losing properties inherent in a picture obtained on interlaced scanning. If an interlaced picture is inputted, and the DCT mode is the field mode, an interlaced picture accommodating picture decoding unit 3 executes 4x4 decimating IDCT. If the DCT mode is the frame mode, the entire coefficients of the DCT block are IDCTed for separation into two pixel blocks associated with interlaced scanning. The separated two pixel blocks are DCTed. The coefficients of the low-frequency components are IDCTed and the two pixel blocks are synthesized together. A progressive picture accommodating picture decoding unit 2 is fed with a progressive picture. The coefficients of the low-frequency components of the DCT block are inverse orthogonal transformed.

What is claimed is:  
     
       1. A picture decoding apparatus for decoding compressed picture data of a first resolution, obtained on predictive coding by motion prediction in terms of a macro-block as a unit and on compression coding by orthogonal transform in terms of an orthogonal transform block as a unit, to obtain moving picture data of a second resolution being lower than the first resolution, comprising: 
       first inverse orthogonal transform means for inverse orthogonal transforming an orthogonal transform block of the compressed picture data, orthogonal transformed by an orthogonal transform system in a field orthogonal transform mode associated with an interlaced scanning;  
       second inverse orthogonal transform means for inverse orthogonal transforming an orthogonal transform block of said compressed picture data, orthogonal transformed in accordance with an orthogonal transform system in a frame orthogonal transform mode associated with a sequential scanning;  
       first addition means for summing the compressed picture data, inverse orthogonal transformed by said first inverse orthogonal transform means or said second inverse orthogonal transform means, to motion compensated reference picture data, to output moving picture data of the second resolution;  
       first memory means for storing moving picture data outputted by said first addition means as reference picture data; and  
       first motion compensation means for motion compensating the macro-block of the reference picture data stored in said first memory means;  
       correction means coupled to the first memory means and receiving picture data therefrom for correcting the picture data by post-filtering for phase deviation between top and bottom fields, with phases of respective pixels of the top field in a vertical direction being (1+2N)/2, where N is a positive even integer including zero and phases of respective pixels of the bottom field in the vertical direction being M, where M is a positive odd integer, so that the corrected phrase of the pixels of the top field in the vertical direction are P, where P is a positive even integer including zero, and so that the corrected phases of the pixels of the bottom field in the vertical direction are Q, where Q is a positive odd integer; said correction means being additionally coupled to the first addition means for converting a picture frame of the first resolution into a picture frame of the second resolution;  
       said first inverse orthogonal transform means inverse orthogonal transforming coefficients of low-frequency components of said orthogonal transform block;  
       said second inverse orthogonal transform means including a first picture decoding unit for inverse orthogonal transforming coefficients of the entire frequency components of said orthogonal transform block, separating respective pixels of the inverse orthogonal transformed orthogonal transform block into two pixel blocks associated with interlaced scanning, orthogonal transforming the separated two pixel blocks, inverse orthogonal transforming the coefficients of the low-frequency components of the two orthogonal transformed pixel blocks and synthesizing the inverse orthogonal transformed two pixel blocks to generate an orthogonal transform block;  
       third inverse orthogonal transform means for inverse orthogonal transforming the orthogonal transform block of compressed picture data, orthogonal transformed in the frame orthogonal transform mode;  
       second addition means for summing the compressed picture data, inverse orthogonal transformed by said third inverse orthogonal transform means, to motion compensated reference picture data, to output moving picture data of the second resolution;  
       second memory means for storing the moving picture data outputted by said second addition means as reference picture data; and  
       second motion compensation means for motion compensating the macro-block of reference picture data stored in said second memory means;  
       said first inverse orthogonal transform means including a second picture decoding unit for inverse orthogonal transforming coefficients of low-frequency components of said orthogonal transform block;  
       conversion means couple to the second memory means and receiving picture data therefrom for converting a picture frame of the first resolution into a picture frame of the second resolution;  
       wherein if compressed picture data of the first resolution, encoded from the moving picture signals of the interlaced scanning system, are inputted, the moving picture data of the second resolution are decoded by said first picture decoding unit; and  
       if compressed picture data of the first resolution, encoded from the moving picture signals of the sequential scanning system, are inputted, the moving picture data of the second resolution are decoded by said second picture decoding unit,  
       whereby with said picture decoding apparatus for decoding compressed picture data of a first resolution, moving picture data of a second resolution lower than the first resolution is obtained regardless of whether the first or the second decoding unit is used.  
     
     
       2. A picture decoding apparatus for decoding moving picture data of a second resolution from compressed picture data of a first resolution, obtained on predictive coding by motion prediction in terms of a macro-block as a unit and on compression coding by orthogonal transform in terms of an orthogonal transform block as a unit, the second resolution being lower than the first resolution, comprising: 
       first inverse orthogonal transform means for inverse orthogonal transforming an orthogonal transform block of the compressed picture data, orthogonal transformed by an orthogonal transform system associated with an interlaced scanning;  
       second inverse orthogonal transform means for inverse orthogonal transforming an orthogonal transform block of said compressed picture data, orthogonal transformed in accordance with an orthogonal transform system associated with a sequential scanning;  
       addition means for summing the compressed picture data, inverse orthogonal transformed by said first inverse orthogonal transform means or said second inverse orthogonal transform means, to motion compensated reference picture data, to output moving picture data of the second resolution;  
       memory means for storing moving picture data outputted by said addition means as reference picture data; and  
       motion compensation means for motion compensating the macro-block of reference picture data stored in said memory means;  
       said first inverse orthogonal transform means inverse orthogonal transforming coefficients of low-frequency components of said orthogonal transform block, correcting the phase by ¼ pixel with respect to a vertical direction of the respective pixels of the top field obtained on inverse orthogonal transform, correcting the phase by ¾ pixel with respect to the vertical direction of the respective pixels of the top field obtained on inverse orthogonal transform;  
       said second inverse orthogonal transform means including a first picture decoding unit for inverse orthogonal transforming coefficients of the entire frequency components of said orthogonal transform block, separating the inverse orthogonal transformed orthogonal transform block into two pixel blocks associated with the interlaced scanning, orthogonal transforming the separated two pixel blocks, inverse orthogonal transforming the coefficients of the low-frequency components of the orthogonal transformed two pixel blocks, correcting the phase of the respective pixels of the top field resulting from inverse orthogonal transform by ¼ pixel, correcting the phase of the respective pixels of the bottom field resulting from inverse orthogonal transform by ¾ pixel, and synthesizing the phase-corrected top and bottom fields; third inverse orthogonal transform means inverse orthogonal transforming an orthogonal transform block of said compressed picture data, orthogonal transformed in the frame orthogonal transform mode;  
       addition means for summing compressed image data inverse orthogonal transformed by said third inverse orthogonal transform means to motion compensated reference picture data to output moving picture data of second resolution;  
       memory means for storing the moving picture data outputted by said addition means as reference picture data; and  
       motion compensation means for motion compensating a macro-block of the reference picture data stored in said memory means;  
       said first inverse orthogonal transform means including a second picture decoding unit for inverse orthogonal transforming the coefficients of low-frequency components of said orthogonal transform block;  
       wherein  
       if compressed picture data of the first resolution, encoded from the moving picture signals of the interlaced scanning system, are inputted, the moving picture data of the second resolution are decoded by said first picture decoding unit; and  
       if compressed picture data of the first resolution, encoded from the moving picture signals of the sequential scanning system, are inputted, the moving picture data of the second resolution are decoded by said second picture decoding unit.  
     
     
       3. A picture decoding method for decoding compressed picture data of a first resolution, obtained on predictive coding by motion prediction in terms of a macro-block as a unit and on compression coding by orthogonal transform in terms of an orthogonal transform block as a unit, to obtain moving picture data of a second resolution being lower than the first resolution, wherein 
       if compressed picture data of the first resolution, encoded from moving picture signals of the an interlaced scanning system are inputted,  
       an orthogonal transform block of the compressed picture data, orthogonal transformed in accordance with the orthogonal transform system associated with the g sequential scanning in a field orthogonal transform mode, is inverse orthogonal transformed;  
       an orthogonal transform block of the compressed picture data, orthogonal transformed in accordance with the orthogonal transform system associated with the sequential scanning in a frame orthogonal transform mode, is inverse orthogonal transformed;  
       the inverse orthogonal transformed compressed picture data is summed to motion compensated reference picture data;  
       reference picture data obtained on addition is stored as reference picture data;  
       a macro-block of the stored reference picture data is motion compensated;  
       coefficients of low-frequency components of the orthogonal transform block, orthogonal transformed in the field orthogonal transform mode, are inverse orthogonal transformed;  
       coefficients of the entire frequency components of said orthogonal transform block, orthogonal transformed in the frame orthogonal transform mode, are inverse orthogonal transformed, pixels of the inverse orthogonal transformed orthogonal transform block are separated into two pixel blocks associated with interlaced scanning, the separated two pixel blocks are respectively orthogonal transformed, coefficients of the low-frequency components of the two orthogonal transformed pixel blocks are inverse orthogonal transformed, and the two inverse orthogonal transformed pixel blocks are synthesized to generate an orthogonal transform block; and  
       the picture data is corrected by post-filtering for phase deviation between top and bottom fields, with phases of respective pixels of the top field in a vertical direction being (1+2N)/2, where N is a positive even integer including zero, and phases of respective pixels of the bottom field in the vertical direction being M, where M is a positive odd integer, so that the corrected phases of the pixels of the top field in the vertical direction are P, where P is a positive even integer including zero, and so that the corrected phases of the pixels of the bottom field in the vertical direction are Q, where Q is a positive odd integer; and, a picture frame of the first resolution is converted into a picture frame of the second resolution;  
       and wherein if compressed picture data of the first resolution, encoded from the moving picture signals of the sequential scanning system, are inputted,  
       the orthogonal transformed block of the compressed picture data, orthogonal transformed in the frame orthogonal transform mode, is inverse orthogonal transformed;  
       the inverse orthogonal transformed compressed picture data and the motion compensated reference picture data are summed together;  
       moving picture data resulting from the summation are stored as reference picture dala;  
       a macro-block of the stored reference picture data is motion compensated;  
       coefficients of the low-frequency components of said orthogonal transform block are inverse orthogonal transformed, and  
       a picture frame of the first resolution is converted into a picture frame of the second resolution;  
       whereby with the picture decoding method for decoding compressed picture data of a first resolution, moving picture data of a second resolution lower than the first resolution is obtained when one of (i) compressed picture data of the first resolution, encoded from moving picture signals of the interlaced scanning system, and (ii) compressed picture data of the first resolution, encoded from the moving picture signals of the sequential scanning system, are inputted.  
     
     
       4. Picture decoding method for decoding moving picture data of a second resolution from compressed picture data of a first resolution, obtained on predictive coding by motion prediction in terms of a macro-block as a unit and on compression coding by orthogonal transform in terms of an orthogonal transform block as a unit, the second resolution being lower than the first resolution, wherein 
       if compressed picture data of the first resolution, encoded from moving picture signals of the an interlaced scanning system are inputted,  
       an orthogonal transform block of the compressed picture data, orthogonal transformed in accordance with the orthogonal transform system associated with a sequential scanning is inverse orthogonal transformed;  
       the inverse orthogonal transformed compressed picture data is summed to motion compensated reference picture data;  
       reference picture data obtained on addition is stored as reference picture data;  
       a macro-block of the stored reference picture data is motion compensated;  
       coefficients of low-frequency components of the orthogonal transform block, obtained on orthogonal transform in the field orthogonal transform mode, are inverse orthogonal transformed, the respective pixels of the top field obtained on inverse orthogonal transform are phase-corrected by ¼ pixel in the a vertical direction of the pixel, the respective pixels of the bottom field obtained on inverse orthogonal transform are phase-corrected by ¾ pixel in the vertical direction of the pixel,  
       coefficients of the low-frequency components of the entire frequency components of the orthogonal transform block, obtained on orthogonal transform in the frame orthogonal transform mode, are inverse orthogonal transformed, the inverse orthogonal transformed orthogonal transform block is separated into two pixel blocks associated with interlaced scanning, the separated two pixel blocks are respectively orthogonal transformed, coefficients of the low-frequency components of the orthogonal transformed two pixel blocks are inverse orthogonal transformed, the respective pixels of the top field obtained on inverse orthogonal transform are corrected for phase by ¼ pixel in the vertical direction of the pixel, die respective pixels of the bottom field obtained on inverse orthogonal transform are corrected for phase by ¾ pixel in the vertical direction of the pixel, and the phase-corrected top and bottom fields are synthesized;  
       and wherein  
       if compressed picture data of the first resolution, encoded from the moving picture signals of the sequential scanning system, are inputted,  
       the orthogonal transformed block of the compressed picture data, orthogonal transformed in the frame orthogonal transform mode, is inverse orthogonal transformed;  
       the inverse orthogonal transformed compressed picture data and the motion compensated reference picture data are summed together;  
       moving picture data resulting from the summation are stored as reference picture data;  
       a macro-block of the stored reference picture data is motion compensated; and  
       coefficients of the low-frequency components of said orthogonal transform block are inverse orthogonal transformed.